[Pw_forum] Use of mixing_beta

[Dr. Robert Molt Jr.]

Good evening,

I have 2 questions about mixing_beta.

1.) The manual states that mixing_beta is a mixing factor for 
self-consistency. Can someone give a reference to how this is defined, 
mathematically?

2.) I have been able to divine from the forums that it is effectively a 
scaling of of the previous SCF cycle's density 
(http://www.mail-archive.com/pw_forum@pwscf.org/msg01302.html) as it 
relates to the current step's calculation. I appreciate that it improves 
SCF convergence.

Broadly, I am confused why this would be useful to speed up SCF 
convergence. I have written my share of SCF codes for molecules with 
basis functions of (gaussians+spherical harmonics), but am new to plane 
waves and PBCs. In molecular calculations, one never scales the previous 
step's density in an iterative solution (or in solving non-linear 
coupled cluster amplitudes). Is this somehow akin to DIIS histories, in 
some way, to be selective in the previous cycle's estimate?

Thank you for your time,

Dr. Robert Molt
Indiana-University Perdue University
ENSCO